Wednesday, June 06, 2007

SoCG 2007: "Magic Hausdorff Lions Have Happy Endings"

(It's at the point now where people complain if the business meeting post is NOT up 5 minutes after the actual meeting. Sigh...)

Summary for people in a hurry:
  • Attendance this year was a record low
  • Next PC Chair: Monique Teillaud
  • SoCG 2008 will be at the University of Maryland (rather than Virginia: long story)
  • SoCG 2009 (by a 4-1 margin) will be in Aarhus, Denmark. BEER !!!!!!!!!
And now for the details:

Local News:
Otfriend Cheong was in charge of local arrangements. I have to say that the hotel we are at is quite amazing: It's on the Bomun Lake, and right outside the hotel is this lake-side path that leads to all the restaurants one can imagine, a tourist village with acres of motorized scooters (!), and some not-disgusting coffee. The hotel facilities themselves are excellent (again, modulo the coffee); much better for a cheaper price in comparison to a US Hotel. And as I sit here writing this, I can hear the rehearsals for our Korean classical music concert tonight.

Unfortunately, there was no one here to enjoy it ! Attendance is down tremendously (111 people), a factor exacerbated by FCRC, being held in a day or two in the exotic locale of San Diego (fish tacos ! blech!), and other related conferences. The relative remoteness of Gyeongju played no small role in this, I imagine.

There was much discussion about whether we should continue to be sponsored by ACM or not (the main issue is cost, and logistics when organizing in a non-US location); no resolution on this point.

PC Stuff: (obvious caveat: I was on the PC this year)
Jeff Erickson presented the usual stats (45/139 papers accepted, out of 286 submitting authors, and with 115+ external reviews). It turns out that the formula for succes at SoCG this year involves
submitting seven papers, with 4 co-authors, from an email address in Pakistan.
The title of this post was composed from words in accepted papers.

Lots of other stats, nothing particularly enlightening.

The Near Future.
Monique Teillaud will chair the SoCG 2008 program committee. The committee has 17 people on it, an unusually large number. She's hoping to get 4 reviews per paper, so maybe that's why.

David Mount explained why we had to move from Alexandria, Virginia to UMD for SoCG 2008. Apparently the main hotel we would have used was bought out and is now much more expensive. Boo hoo. On the bright side, UMD will be a lot cheaper in terms of accomodation.

SoCG 2009.

We had two competing bids for SoCG 2009. Aarhus (Beer ! Lego ! Beer ! City life ! Beer!) and Portland (Roses ! Beer ! More Roses ! Powell's Books ! Microbreweries!).

No, we are not a bunch of boozing alcoholics.

Lars did a bang up job with his Aarhus presentation. John Hershberger did a great presentation on Portland (a great city to visit, btw), but it was no contest. By a 4-1 margin, Aarhus it is !

Open Discussion.
It was getting rather late by the time we got to open discussions. Guenter Rote initiated the question, "Should we have a rebuttal process for reviewing papers", in response to apparently some aggrieved set of rejected authors.

We had a heated discussion on this point, and although we ultimately went nowhere, it was felt that we should continue things electronically (i.e on blogs). I'll post at greater length on this issue later, but to summarize the main points pro and con:

  • Creates an improved sense of fairness
  • A safety net for potential errors
  • Time waste for reviewers
  • Time waste for authors (!) (according to Jack Snoeyink), but I am not convinced that this is a valid argument
  • Won't make a significant difference
I'm personally dubious whether there's any measurable benefit to a rebuttal process, but I'm mildly in favor primarily because of the "placebo effect" it has.


  1. I'm not a big fan of rebuttal phases. They can lead to more frustration for authors: "I gave a perfect explanation of why the reviewers are idiots and *still* they rejected my paper". Consequently, I do think they can count as a waste of time for authors. Preparing a rebuttal can easily take up a full day, more if you count mentally drafting responses in the shower etc. So not always placebo, possibly added irritant.


  2. Gosh. Nothing wrong with having SoCG in Korea, but why

    i) so far away from Seoul,

    ii) the same year as FCRC, and

    iii) in such a close proximity in time to FCRC?

    Does the SoCG steering committee has a death wish? Or is this just a way to keep the CG small? or was it just a momentary lapse in judgement?

  3. Rebutals have a useful placebo effect in that totally inane comments can be rebutted. We all have at one point or another received a comment saying "the authors should have compared their work with x" or "this is superseeded by x" when in fact the paper does compare with x or explicitly explains why x does not superseed the work, but the reviewer missed it, under the tremendous time pressures they are normally under.

    Those clear, horrible errors get taken care of during the rebutal phase and presumably any remaining reasons for rejection are now more subjective and harder to take issue with even if one ultimately disagrees.

  4. I think a rebuttal phase would *save* authors' time. Here's a typical scenario of what happens without a rebuttal phase: a referee writes a totally inane comment, and the paper gets rejected. Now the authors have to submit their paper elsewhere, and have to take into account that there is at least one referee and probably a couple of PC members out there who believe this inane comment is correct. Since our community is relatively small, it is quite likely that the paper ends up being refereed by one of these people again. So the authors end up spending a lot of effort to rewrite their paper to rebut one inane comment by a referee who was not paying attention. This is a much bigger waste of time then writing a short mail to the PC to rebut the inane comment when it pops up.

    On a related note, my main problem with refereeing processes is that the authors get no insight in why the PC took the decision they took. It would help to understand and accept the decisions taken by PC's if the authors get at least the same information about their paper which the PC gets, namely the *full* referee reports. I just got a paper rejected from another conference with two out of three referee reports empty, that is, as far as the authors get to see them. I think this is outrageous; next time I am asked to referee for that particular conference I may consider to simply return the empty referee reports for reuse.

    How about we completely abandon the concept of "comments to the PC (not sent to the authors)"? Actually as a referee I never really understood why it was there in the first place: it can be convenient but I can do without.

  5. It strikes me that in the distant past, PC chairs were supposed to go over reviewer comments. In the process they would rap reviewers for inappropriate comments and fix things such as confidentiality leaks.

    With the way things are now, with very large conferences still being led by a single PC chair, we can no longer expect this type of quality control. Perhaps conferences should appoint vice-chairs who are responsible for reading reviewer comments, making sure every reviewer wrote comments to the authors, reflecting the score given, not leaking confidential info, etc.

  6. This is already happening in some large conferences, with the idea of "senior reviewers" and "junior reviewers". The latter have traditional reviewing responsibilities, and the senior reviewers are assigned disjoint subsets of papers to oversee.

    One problem with this approach is that the "junior PC members" end up acting like external reviewers. which is to say, they don't continue to have a global view of the entire catalog of submissions: this has to be left up to to the senior PC members


Disqus for The Geomblog